Journal of Applied Mathematics and Computational Mechanics 2014, 13(2), 63-71

DIFFERENTIATION AND INTEGRATION

BY USING MATRIX INVERSION

Dagmara Matlak, Jarosław Matlak, Damian Słota, Roman Wituła

Institute of Mathematics, Silesian University of Technology, Gliwice, Poland

roman.witula@polsl.pl

Abstract. In the paper certain examples of applications of the matrix inverses for generat-
ing and calculating the integrals are presented.

Keywords: matrix inverse, integrals, generalized McLaurin’s formula

Introduction

The first part of our discussion concerns the linear mappings defined on the
finite-dimensional space of solutions of the following system of differential equa-
tions

 = 
,

+ ⋯ + 
, 

 = ,

+ ⋯ + , 

(1)

   ⋯ + ,  .
⋮

= ,

+

Suppose that functions 
,  , … ,  form the solution of the above system of

equations and matrix  = , × of system (1) is nonsingular. Let us consider
the linear mapping  of the linear space of solutions 
,  , … ,   of system (1)
onto itself defined in the following way:



 
,

+ ⋯ + 
,  
,

, 


 ⋮  =  ⋮  =  = ⋮ ⋮   ⋮  =   ⋮ . (2)
…

  ,

+ ⋯ + ,  ,
,  

⋮ ⋮
…

Matrix  is nonsingular, so there exists its inverse 
. In particular, the follow-
ing equality occurs:



    =   ⋮ .
 

⋮ 

(3)
64 D. Matlak, J. Matlak, D. Słota, R. Wituła




Therefore, if    =  ⋮ , then functions  are the primitive functions
 

⋮

of  ( = 1,2, … , ).
The current article is inspired by Swartz’s paper [1] where the author gives

integrals of functions ℎ
= e
sin , ℎ = 
cos , for which he obtained the
some simple examples of using this procedure, among others, for generating the

formula (the integration constants are omitted and this rule will oblige hencefor-
ward):

 ℎ    #  −  $%# 

=  ! = మ  మ " &.
 ℎ    ℎ  #  +  $%# 
− ℎ
 ೌೣ
(4)

1. Generalization of Swartz’s example

Let us start from the generalization of the example mentioned above. Let us take
the functions


 = cosh  sin  ,   = cosh  cos  ,

  = sinh  cos  ,   = sinh  sin  .
(5)

Note that the differentiation operator for these functions is of the form



   

   
 ' ( = *  , = - . ' (.
0 0

  − 

 − 0 0 

 
(6)
) +  
 0 0
 0 0


If  ≠ 0, then the inverse of matrix of operator  has the form


 

= - ..
0 − 0

 

0 0
(7)

మ  మ 0 0
0 − 0

Now we can easily integrate functions  , e.g.

−  +  
/ 
  =
 +  
.
 Differentiation and integration by using matrix inversion 65

2. Integrals of ‫ ࢞ ࢔ܖܑܛ‬for odd ࢔

Consider the second derivatives of functions  = sin  ,  ≥ 2. We have

sin   =  sin
 cos   =  − 1 sin  −  sin . (8)
Of course for  = 1 there is sin   = − sin . Thus we can write the second
derivative operator for the odd powers of function sin , from 1 to odd 0, in the
following matrix form:

sin  sin  − sin 

sin  6 sin  − 9 sin 


sin 
 ' (=* ,=' (=
  

sin  )1sin 2 + 00 − 1 sin  − 0  sin 

⋮ ⋮ ⋮



sin 
(9)

sin 
−1
=- .' (.
0 ⋯ 0 0

3 ∙ 2 −3 ⋯ 0 0

⋯ 00 − 1 −0  sin 
⋮ ⋮ ⋱ ⋮ ⋮ ⋮
0 0

The determinant of the obtained matrix is equal to: det  = (−1)

ೖశభ

0 ∈ 3. The inverse matrix is of the form

మ (0‼) ≠ 0,



 = , ೖశభ×ೖశభ , (10)
మ మ

where
 < 4,

 = 4,
0,

, =

 
‼
‼
− 
మ ,

 > 4.
(11)

− ‼ 
 ‼ ,
We can deduce that for odd  there occurs (with respect to the linear element):

 sin   = − ∑ sin
 ,

 
‼ 
⁄
(
)‼
(12)
‼ 
()‼

where   : =    . For example, we get



 sin   = −
 sin  −  sin  −  sin  .

  

(13)

We note that from (12) by differentiating we obtain (see [2, 3]):

 sin   = − cos  ∑ sin  .


‼ 
⁄ (
)‼
(14)
‼ ()‼
66 D. Matlak, J. Matlak, D. Słota, R. Wituła

For example, we have

 sin   = −
 cos  (1 +  sin  +  sin ).




3. The case of the even powers of ‫࢞ ܖܑܛ‬

  = sin  − ೙షభ sin ,

Consider functions of the form
(15)
 
೙

 
for  = 2,4,6, … . Acting on the vector ' (, where 0 is even, with the second
 
⋮

derivative operator, like it was done in equation (9), we get the following transfor-

    
mation matrix:

    
6 '  ( = '  (=
    
⋮ ⋮
(16)
 
2 ∙ 3⁄4 −4 …  


=- .'  (.
−2 0 … 0 0
0 0

… 0 − 2  0 − 1 ⁄0 −0   
⋮ ⋮ ⋱ ⋮ ⋮ ⋮
0 0
The above matrix is invertible and its inversion is of the form
6
= , ೖ×ೖ, (17)

 < 4,
మ మ


 = 4,
0,

, =


− మ ,

 > 4.
where (18)

− మ 
‼ ‼ ,


‼ ‼

Therefore for even n we get the formula (exact to the linear element):

    ∑
"sin  − sin &

 
‼ ⁄ ‼ 

= −

" sin  − 1 +
‼మ 
‼ 

‼ ‼
= −

+ ∑

sin  
‼ − 
‼ &

‼మ 
‼
‼ ‼ 

(19)
⁄

− మ sin  = − మ sin  ,

‼


= ‼ మ
 Differentiation and integration by using matrix inversion 67

which implies the following integral identity

 sin   = "  + ∑
   &

 
‼   ⁄ ‼ 

" − ∑
& .

‼ (
)‼

‼
మ ‼
(20)
⁄ 
= ‼  (
)‼() మ sin

Hence, by differentiating we get (see [2, 3]):

 sin   = " − cos  ∑
sin
&.


‼  ⁄ ‼
(21)
‼ (
)‼

For example, we obtain

 sin   =   − cos  "sin  +  sin  +
 sin &!.

  
(22)

4. Integral of ‫࢞ ࢔ܖ܉ܜ‬

Let 8 be the linear space of sequences 9  :

 : (, ) → ;. Let <: 8 → 8 be a linear operator satisfying equation
 of differentiable functions

tan  
tan  tan 
' ( = < '  (.
1

tan   tan 
 
(23)
⋮ ⋮
If < is represented by infinite matrix , then from (23) matrix  has the form

=- ..
1 0 1 0 0 ⋮
0 2 0 2 0 ⋮
(24)
0 0 3 0 3 ⋮
⋯ ⋯ ⋯ ⋯ ⋯ ⋯
It is easy to show that

?1 0 − 0 ⋮B

> ⋮ A.
0


= >0  A

>0 0 0 − ⋮ A
0 − 0 (25)

=⋯ ⋯ ⋯ ⋯ ⋯ ⋯@





However, matrix 
does not represent the inverse operator <
, since the fol-
lowing relations hold
68 D. Matlak, J. Matlak, D. Słota, R. Wituła

tan  
tan   tan 
' (=- . '  (, (26)
1 0 0 0 −1 0 0 0 ⋮ 1

tan  tan 
0 1 0 0 0 −1 0 0 ⋮
  0 0 1 0 0 0 −1 0 ⋮
⋮ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋯ ⋮
resulting from the formula

" tan
 tan & = tan  − tan .




− (27)




Moreover, we get from this, after summing over powers in the interval − ,  and
 
 
by uniform convergence (see [4]), that

tan  = 
∑
 "
tan  −  tan &,




(28)

or equivalently

 tan CC = ∑
 " tan
 tan &




−

= ∑ ,



ೖ
(29)


 
tan

for every  ∈ −  ,   and  = 0,1,2, … .

 

From formula (28) we also get the matrix form D< of operator <
, i.e. the
inverse operator of operator < (under assumption of its existence):

? B
1 0 0 0 0 ⋯
> 0 ⋯A

>−
0 0 ⋯A
0 
0 0

D< = > A.


0
>0 − 0 ⋯A
 
(30)
>
A



0 

>  A


= ⋮ ⋮ ⋱@
0 − 0 
⋯
⋮ ⋮ ⋮
Formula (29) is our main analytic result in this section. Why do we think so?
Because, as we show now, this formula is a generalization of the classical
MacLaurin’s formulae for ln( + 1), i.e.

ln + 1 =  −

మ
య
ర
+ − + ⋯, (31)
for −1 <  ≤ 1, found independently by Nicolaus Mercator and Saint-Vincent
  

(see sections 10-9 and 10-10 in [5] and page 387 in [6]), and for arctan , i.e.

arctan  =  −

య
ఱ
+ − ⋯, (32)
for −1 <  ≤ 1, which is known as the Gregory series.
 
 Differentiation and integration by using matrix inversion 69

This connection should not be surprising because of the known complex rela-
tion (see [7]):

arctan E = ln " &,


(33)

where E⁄ ∈
 

(−∞, −1] ∪ [1, ∞) and where the principal branch of the loga-
rithm is under consideration. On the cuts we have

arctan(C) = ± + ln " &,

  

(34)
  

for C ∈ (−∞, −1) ∪ (1, ∞) and where the upper/lower sign corresponds to the
right/left side of the set determining C. More precisely, the connection between the

imaginary parts of arctan E, since we have

arctan function and log function is obvious and the section concerns the real and

arctan E =  arctan "


మ  మ & +   ln "
మ 
మ &,




మ 
మ
(35)

where E =  + C, |E| < 1. First, from equation (29) for  = 1 we get

− ln cos  = ∑ 

ೖ
(
)
 
 tan (36)

or

ln cos1arctan √2 =  ∑


ೖ


, (37)

which by (31) implies the well known identity (since cos H =

for
H ∈ "−  ,  &):
√
 !"మ #
 

ln cos1arctan √2 = −  ln + 1 ,


(38)

i.e.

√ + 1 cosarctan  ≡ 1, (39)

for every  ∈ I0,1J.
But this formula holds for every  ≥ 0 since cos H =


for every
H∈ "−  ,  &. In other words, formula (37) is equivalent to (31). For  = 0 from
√
 !"మ #
 

(29) we get

= ∑
ೖ
!$% !"
&
మ '
 , (40)



70 D. Matlak, J. Matlak, D. Słota, R. Wituła

which implies (32). Hence, for  = 1 we obtain





=1− + − +⋯ (41)
   (

That is the classical Gregory-Leibniz-Nilakantha’s formula (see [8]). Generaliza-

As seen from equation (29), the values of integrals ర tan   for  ≥ 2 are
tions of the Gregory power series (32) are discussed in papers [9] and [10].
ഏ

the translations of numbers ర tan   or ర , depending on parity of . For
ഏ ഏ

even  we have

ర tan   = ర  − ∑ =  − ∑

ഏ ഏ
()⁄ (
)ೖ  ()⁄ (
)ೖ
, (42)

whereas for odd  we get




ర tan   = ర tan   − ∑
=  "ln 2 − ∑
&.(43)

ഏ ഏ
(
)⁄ (
)ೖ
(
)⁄ (
)ೖ
 

5. Final remark

Some other applications of the matrix obtained by n-times differentiation of

product functions and composition functions are discussed in paper [11]. In turn, in

 sec 
 , similarly as in the present study. The obtained formulae were used
paper [12] the technique of the inverse matrix was used for calculating the integral

there for generating the trigonometric identities.

References
[1] Swartz W., Integration by matrix inversion, Amer. Math. Monthly 1958, 65, 282-283.
[2] Prudnikov A.P., Bryczkov J.A., Mariczev O.J., Integrals and Series, Elementary Functions,
Vol. 1, Nauka Press, Moscov 1986 (in Russian).
[3] Prudnikov A.P., Bryczkov J.A., Mariczev O.J., Integrals and Series, Complements Sections,
Vol. 2, Nauka Press, Moscow 1986 (in Russian).
[4] Kołodziej W., Mathematical Analysis, PWN, Warsaw 1979 (in Polish).
[5] Eves H., An Introduction to the History of Mathematics, Holt, Rinehart and Winston, New York
1969.
[6] Boyer C. (revised by Merzbach U.), A History of Mathematics, Wiley, New York 1991.
[7] Olver F.W.J., Lozier D.W., Boisvert R.F., Clark C.W., NIST Handbook of Mathematical Func-
tions, Cambridge Univ. Press, Cambridge 2010.
[8] Wituła R., Number π, its History and Influence on Mathematical Creativeness, Wyd. Pracowni
Komputerowej Jacka Skalmierskiego, Gliwice 2011 (in Polish).
 Differentiation and integration by using matrix inversion 71

[9] Gawrońska N., Słota D., Wituła R., Zielonka A., Some generalizations of Gregory’s power
series and their applications, J. Appl. Math. Comput. Mech. 2013, 12(3), 79-91.
[10] Wituła R., Hetmaniok E., Pleszczyński M., Słota D., Generalized Gregory’s series (in review).
[11] Redheffer R., Induced transformations of the derivative - vector, Amer. Math. Monthly 1976,
83, 255-259.
[12] Wituła R., Matlak D., Matlak J., Słota D., Use of matrices in evaluation of ‫ ׬‬sec ଶ௡ାଵ ‫ݔ݀ ݔ‬
in review.